Fig. 4: Performance evaluation in real displacement data.
From: Estimating full-field displacement in biological images using deep learning
Visualisation of manually tracked displacements in Drosophila confocal microscopy videos of (a) F-actin, GMA-GFP (used in the training set) and (b) myosin, Sqh::GFP (unseen by training). Scale bars are 20 μm. Representative displacement tracks in (c) F-actin and (d) myosin datasets evaluated using conventional methods compared to DEFORM-Net. Displacement error compared to manual tracking in (e) F-actin (n = 9) and (f) myosin (n = 9) datasets. Error bars are standard error of mean.
